| 205 | 0 | 58 |
| 下载次数 | 被引频次 | 阅读次数 |
【目的】准确的水位预测对防洪减灾、航运安全和水资源管理至关重要。传统水文模型往往难以捕捉水位波动中复杂的时空依赖关系和非线性特征。【方法】针对这些局限性，提出HydroFusionNet深度学习框架，通过并行处理CNN-LSTM模块与自注意力机制提升预测精度：CNN-LSTM模块提取局部时空特征，自注意力机制捕获全局上下文关联，最终融合二者输出结果。该框架基于昇腾CANN架构优化部署，利用高性能计算实现高效训练与预测。【结果】在西江流域数据集上的实验表明，HydroFusionNet的RMSE指标达0.394,R2达0.894,性能优于传统模型。昇腾CANN架构使其训练速度提升2.3倍，能耗降低17%,展现出显著计算优势。【结论】该模型能有效刻画复杂水文动态特征，为智能水文管理与灾害防控提供实时水位预测支持。
Abstract：[Objective] Accurate water level prediction is essential for flood control, navigation safety, and water resource management. Traditional hydrological models often fail to capture the complex spatiotemporal dependencies and nonlinear characteristics in water level fluctuations. [Methods]To address these limitations, the Hydro Fusion Net deep learning framework was proposed, integrating CNN-LSTM module and self-attention mechanism in a parallel architecture to improve prediction accuracy. The CNN-LSTM module extracted local spatiotemporal dependencies, while the self-attention mechanism captured global contextual relationships. The outputs of both modules were fused for final prediction. This framework was optimized for deployment on the Ascend CANN platform, leveraging high-performance computing to achieve efficient training and prediction.[Results]Experiments on the Xijiang River dataset showed that Hydro Fusion Net outperformed conventional models, achieving an RMSE of 0. 394 and an R2 of 0. 894. Furthermore, with optimization on Ascend CANN, the training speed increased by 2. 3 times, and energy consumption reduced by 17%, indicating significant computational advantages. [Conclusion] The proposed framework effectively captures complex hydrological dynamics, providing real-time water level prediction support for intelligent hydrological management and disaster prevention.
[1]XIANG X J,LI Q,KHAN S,et al.Urban water resource management for sustainable environment planning using artificial intelligence techniques[J].Environmental Impact Assessment Review,2021,86:106515.
[2]JI X Y,SUN Z B,LV H F,et al.Spatio-temporal multivariable time vario-zoom network for water level forecasting based on high-resolution hydrological dataset[J].Journal of Hydrology,2024,634:131060.
[3]BARRIENTOS-TORRES D,MARTINEZ-RÍOS E A,NAVARRO-TUCH S A,et al.Water flow modeling and forecast in a water branch of Mexico City through ARIMA and transfer function models for anomaly detection[J].Water,2023,15(15):2792.
[4]ALI MUSARAT M,ALALOUL W S,ALI RABBANI M B,et al.Kabul river flow prediction using automated ARIMA forecasting:Amachine learning approach[J].Sustainability,2021,13 (19):10720.
[5]AFZAL M A,ALI S,NAZEER A,et al.Flood inundation modeling by integrating HEC-RAS and satellite imagery:A case study of the Indus River Basin[J].Water,2022,14(19):2984.
[6]DESALEGN H,MULU A.Mapping flood inundation areas using GISand HEC-RAS model at Fetam River,Upper Abbay Basin,Ethiopia[J].Scientific African,2021,12:e00834.
[7]JUNG C,AHN S,SHENG Z P,et al.Evaluate river water salinity in semi-arid agricultural watershed by coupling ensemble machine learning technique with SWAT model[J].JAWRA Journal of the American Water Resources Association,2022,58(6):1175-1188.
[8]JANJIC J,TADIC L.Fields of application of SWAT hydrological model:A review[J].Earth,2023,4(2):331-344.
[9]QI W C,MA C,XU H S,et al.A review on applications of urban flood models in flood mitigation strategies[J].Natural Hazards,2021,108(1):31-62.
[10]ZHAN Y,ZHANG H J,LIU Y.Forecast of meteorological and hydrological features based on SVR model[C]//IEEE.2021 4th International Conference on Advanced Electronic Materials,Computers and Software Engineering (AEMCSE).Changsha:IEEE,2021:579-583.
[11]YOUSSEF A M,PRADHAN B,DIKSHIT A,et al.Comparative study of convolutional neural network (CNN) and support vector machine (SVM) for flood susceptibility mapping:A case study at Ras Gharib,Red Sea,Egypt[J].Geocarto International,2022,37(26):11088-11115.
[12]NAIM A,AAROUD A,AKODADI K,et al.A fully AI-based system to automate water meter data collection in Morocco country[J].Array,2021,10:100056.
[13]GHAHFAROKHI M,FARHANG M,SONBOLESTAN S H,et al.Acomparative study of convolutional and recurrent neural networks for storm surge prediction in Tampa Bay[J].ar Xiv preprint,2024:2408.05797.
[14]DING Y K,ZHU Y L,FENG J,et al.Interpretable spatio-temporal attention LSTM model for flood forecasting[J].Neurocomputing,2020,403:348-359.
[15]KHORRAM S,JEHBEZ N.A hybrid CNN-LSTM approach for monthly reservoir inflow forecasting[J].Water Resources Management,2023,37(10):4097-4121.
[16]PAN X R,GE C J,LU R,et al.On the integration of self-attention and convolution[C]//IEEE.2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).New Orleans:IEEE,2022:805-815.
[17]WEI X K,WANG G J,SCHMALZ B,et al.Evaluation of Transformer model and Self-Attention mechanism in the Yangtze River basin runoff prediction[J].Journal of Hydrology:Regional Studies,2023,47:101438.
[18]CASTANGIA M,GRAJALES L M M,ALIBERTI A,et al.Transformer neural networks for interpretable flood forecasting[J].Environmental Modelling&Software,2023,160:105581.
[19]XU J X,FAN H X,LUO M H,et al.Transformer based water level prediction in Poyang Lake,China[J].Water,2023,15(3):576.
[20]SHEN Y C,RUIJSCH J,LU M,et al.Random forests-based errorcorrection of streamflow from a large-scale hydrological model:Using model state variables to estimate error terms[J].Computers&Geosciences,2022,159:105019.
[21]TRÄFF E A,RYDAHL A,KARLSSON S,et al.Simple and efficient GPU accelerated topology optimisation:Codes and applications[J].Computer Methods in Applied Mechanics and Engineering,2023,410:116043.
[22]LI Y,WANG K,ZHENG H,et al.Ascend:A scalable and energyefficient deep neural network accelerator with photonic interconnects[J].IEEE Transactions on Circuits and Systems I:Regular Papers,2022,69(7):2730-2741.
[23]PRABU R,DARVIN V,MONISH C,et al.Disorder detection on EEG signals and brain images using a hybrid approach of combining the convolutional neural network and artificial neural network(CANN)[C]//IEEE.2024 3rd International Conference on Artificial Intelligence For Internet of Things (AIIo T).Vellore:IEEE,2024:1-6.
[24]SUMON M F I,KHAN M A,RAHMAN A.Machine learning for real-time disaster response and recovery in the US[J].International Journal of Machine Learning Research in Cybersecurity and Artificial Intelligence,2023,14(1):700-723.
[25]AVILA A,SANTOS H,CRUZ A,et al.Hybri D-GM:A framework for quantum computing simulation targeted to hybrid parallel architectures[J].Entropy,2023,25(3):503.
[26]VALENTINI M,DOS SANTOS G B,MULLER VIEIRA B.Multiple linear regression analysis (MLR) applied for modeling a new WQIequation for monitoring the water quality of Mirim Lagoon,in the state of Rio Grande do Sul-Brazil[J].SN Applied Sciences,2021,3:1-11.
[27]KOUADRI S,PANDE C B,PANNEERSELVAM B,et al.Prediction of irrigation groundwater quality parameters using ANN,LSTM,and MLR models[J].Environmental Science and Pollution Research,2022,29(14):21067-21091.
[28]PAN M Y,ZHOU H N,CAO J Y,et al.Water level prediction model based on GRU and CNN[J].IEEE Access,2020,8:60090-60100.
[29]WANG X J,SUN R C,DUAN Q Y,et al.Comparison of deep learning and response surface surrogates in calibration of distributed models[J].South-to-North Water Transfers and Water Science&Technology,2025,23(6):1401-1412.
[30]XIAO Y,WANG Z K.AIbench:A tool for benchmarking Huawei ascend AI processors[J].CCF Transactions on High Performance Computing,2024,6(2):115-129.
[31]CHICCO D,WARRENS M J,JURMAN G.The coefficient of determination R-squared is more informative than SMAPE,MAE,MAPE,MSE and RMSE in regression analysis evaluation[J].Peer JComputer Science,2021,7:e623.
[32]LEMAY A,HOEBEL K,BRIDGE C P,et al.Improving the repeatability of deep learning models with Monte Carlo dropout[J].NPJ Digital Medicine,2022,5:174.
基本信息:
DOI：10.13928/j.cnki.wrahe.2026.07.014
中图分类号:P338
引用信息:
[1]施朝昱,吕海峰,苏冬冬,等.HydroFusionNet:面向西江水位预测的CNN-LSTM-自注意力网络及昇腾CANN适配（英文）[J].水利水电技术(中英文),2026,57(07):182-194.DOI:10.13928/j.cnki.wrahe.2026.07.014.
基金信息:
National Natural Science Foundation(62466051); Guangxi Young Teachers' Basic Ability Improvement Program(2024KY0692); Autonomous Region-level College Students' Innovation and Entrepreneurship Training Program(S202411354227)~~
2025-08-05
2025-08-05
2025-08-05